Hi Paco,
Great article.
I like the fact that you refer back to where this stuff comes from. Too often, at least in the computer industry, we think that what was done before is does not matter. In here you reach back to 1962.
We can also mention things like the standard algorithm for K-Means was created in 1957 and the ID3 algorithm for decision trees in 1979. More importantly, we should remember the incredible work done in the early days of computing that led to what we have today.
At the end of the article you say:
"
We found that 23% of the enterprise organizations attempting to leverage data science, machine learning, artificial intelligence, etc., cite recognize business use case as a critically missing skill within their teams. What would you call that role? Where and how does a person learn to perform it?" I don't know how to call that role. Is it really a one person job or, instead, a discussion within a team of people with different expertise. I just went through this last week. We'll see how successful this project will be but I believe we successfully found a business use case.
So, how does a person learn to perform it? I think it is a mix of science and art so parts can be taught but then there is practice and experience. Here are things I think would be useful to learn:
- How to ask questions and follow up to questions
how do the business work, where the money is spent, where it is made, pain points for growth, scalability, etc.
- Keep it simple (how to say no?)
How do we limit the use case? Subdivide it to basic elements that can be tackled separately.
I'm a big advocate of starting small, get some successes and build on them.
- Know what is possible
Understand what can be done with different algorithms (decision trees, K-Means, ALS, etc). How can they be applied creatively (predict, classify, cluster) in the context of business pain points.
(I would start with looking for a "shallow" learning solution first in the spirit of "keep it simple")
- Know available tools
This includes more than knowing about Spark and notebooks for example. What about available models that could be applied without having to create them from scratch?
So, I see this as a team effort where communication is essential between business people and "data science"-oriented people.
I know this is no where close to answering your questions but hopefully this contributes to advancing the discussion.
------------------------------
Jacques Roy
Digital Technical Engagement
Watson Data and AI
see: youtube.com/c/ByteSizeDataScience
------------------------------
Original Message:
Sent: Tue April 02, 2019 07:06 PM
From: Paco Nathan
Subject: "What Is Data Science?"
Here's a discussion thread for the "What Is Data Science?" article.
Glad to hear all feedback, suggestions, etc.
------------------------------
Paco Nathan
------------------------------
#GlobalAIandDataScience
#GlobalDataScience